Pension Plans: Characteristics of Persons in the Labor Force Without
Pension Coverage (Letter Report, 08/22/2000, GAO/HEHS-00-131).

Pursuant to a congressional request, GAO provided information on the:
(1) portion of the labor force without pension coverage and how that
proportion has changed over the past decade; (2) characteristics of
workers in that labor force; and (3) proportion and characteristics of
retired people who lack pension income or pension assets.

GAO noted that: (1) about 53 percent of the employed labor force lacked
a pension plan in 1998, a decrease in those without coverage of 5
percentage points from 10 years earlier; (2) this improvement in pension
coverage may stem from the economic expansion under way since 1991 that
has encouraged firms to offer pensions as a part of their compensation
packages and from an increased interest in pension coverage by persons
in the labor force; (3) about 39 percent of the employed labor force
lacked a pension plan because they worked for firms that did not sponsor
a plan, while 14 percent lacked a plan because they were not eligible or
chose not to participate in their firm's plan; (4) in 1998, about 85
percent of employees not in a firm's plan had one or more of the
following characteristics: (a) they had relatively low income; (b) were
employed part time or part of the year; (c) worked for a relatively
small firm; or (d) were relatively young; (5) 22 percent of all
employees worked for firms that had fewer than 25 employees, and 82
percent of them lacked pension coverage; (6) these characteristics
appear to be associated with an employee's desire for, or ability to
take advantage of, pension coverage and a firm's willingness or ability
to provide coverage; (7) the per capita cost of sponsoring a pension
plan may be higher for smaller firms than for larger firms; (8) GAO's
analysis indicates that in the past, many workers failed to earn a
pension benefit during their work lives; (9) retired people without
pension income were more likely to be single, female, less educated, and
Hispanic or not white; (10) additionally, retired persons who lacked
pension income were more likely to be poor; and (11) about 21 percent of
retired persons without pension income had incomes below the federal
poverty threshold, compared with 3 percent with pension income.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-00-131
     TITLE:  Pension Plans: Characteristics of Persons in the Labor
	     Force Without Pension Coverage
      DATE:  08/22/2000
   SUBJECT:  Retirement pensions
	     Investment planning
	     Employee retirement plans
	     Statistical data
	     Labor force
IDENTIFIER:  BLS Current Population Survey

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO/HEHS-00-131

Appendix I: Scope and Methodology

24

Appendix II: Analysis of CPS Pension Coverage Rates and Lack-of-Coverage
Factors

27

Appendix III: Summary of GAO's Regression Models

30

Table 1: Characteristics of Retired Persons Who Had Incomes Below
the Poverty Threshold and Lacked Pension Income, 1998 22

Table 2: Pension Status of the Wage and Salary Labor Force, 1988
and 1998 27

Table 3: Major Characteristics of Employees Not in a Pension
Plan, 1998 28

Table 4: Characteristics of Retired Persons With No Pension Income
of Their Own or From a Spouse, 1998 29

Table 5: Characteristics Associated With Persons in the Labor Force
Who Lacked Pension Coverage, 1993 31

Table 6: Characteristics Associated With Persons in the Labor Force
Who Chose Not to Participate in a Pension Plan of the 401(k)
Type, 1993 34

Figure 1: The Proportion of the Employed Labor Force That Lacked Pension
Coverage, 1988 and 1998 8

Figure 2: The Proportion of Employees With Selected Characteristics
Who Did Not Have a Pension Plan, 1998 12

Figure 3: Percentage of Retired Persons With and Without Pension
Income Who Had Selected Characteristics, 1998 20

Figure 4: Sources of Income for Retired Persons With and Without
Pension Income, 1998 21

BLS Bureau of Labor Statistics

CPS Current Population Survey

SSI Supplemental Security Income

Health, Education, and
Human Services Division

B-284459

August 30, 2000

The Honorable Major R. Owens
Ranking Minority Member
Subcommittee on Workforce Protections
Committee on Education and the Workforce
House of Representatives

The Honorable Robert Andrews
Ranking Minority Member
Subcommittee on Employer-Employee Relations
Committee on Education and the Workforce
House of Representatives

People who retire rely on income from pensions, Social Security, personal
savings, and--to the extent they are willing and able to continue
working--wages and salaries.1 The dramatic and continued growth in the size
of the nation's elderly population, and the resulting stress on Social
Security's finances, may portend greater reliance on other resources, such
as pensions, for the economic well-being of people who retire.

In recent decades, the Congress has enacted legislation to encourage pension
sponsorship and participation. However, questions remain about the portion
of the labor force not covered by pension plans and retired people who lack
pension income. You asked us to provide information concerning (1) the
proportion of the labor force without pension coverage and how that
proportion has changed over the past decade, (2) the characteristics of
workers in that labor force, and (3) the proportion and characteristics of
retired people who lack pension income or pension assets.

To provide this information, we analyzed data from the Current Population
Survey (CPS), conducted by the Bureau of the Census for the Bureau of Labor
Statistics (BLS). For our analysis, we developed two regression models based
on CPS data that examine the association between various demographic and
other characteristics and the lack of pension coverage. We also reviewed
studies and talked to labor experts in government, the private sector, and
academia to collect information on the reasons why certain characteristics
are associated with the lack of coverage. We conducted our work between
August 1999 and April 2000 in accordance with generally accepted government
auditing standards. See appendix I for our scope and methodology.

About 53 percent of the employed labor force lacked a pension plan in 1998,
a decrease in those without coverage of 5 percentage points from 10 years
earlier. This improvement in pension coverage may stem from the economic
expansion under way since 1991 that has encouraged firms to offer pensions
as a part of their compensation packages and from an increased interest in
pension coverage by persons in the labor force. About 39 percent of the
employed labor force lacked a pension plan because they worked for firms
that did not sponsor a plan, while 14 percent lacked a plan because they
were not eligible or chose not to participate in their firm's plan.

A few characteristics describe the large majority of persons in the employed
labor force who do not have a pension plan. In 1998, about 85 percent of
employees not in a firm's plan had one or more of the following
characteristics: They had relatively low income, were employed part time or
part of the year, worked for a relatively small firm, or were relatively
young. In that year, for example, 40 percent of all employees earned less
than $20,000, and 81 percent of them lacked pension coverage. Similarly, 22
percent of all employees worked for firms that had fewer than 25 employees,
and 82 percent of them lacked pension coverage. These characteristics appear
to be associated with an employee's desire for, or ability to take advantage
of, pension coverage and a firm's willingness or ability to provide
coverage. For example, persons earning less than $20,000 per year may have
insufficient income to both pay current expenses and save for retirement.
The per capita cost of sponsoring a pension plan may be higher for smaller
firms than for larger firms.

Our analysis indicates that in the past, many workers failed to earn a
pension benefit during their work lives. In 1998, for example, about 48
percent of persons who had retired lacked pension income or annuities.
Retired people without pension income were more likely to be single, female,
less educated, and Hispanic or not white. Additionally, retired persons who
lacked pension income were more likely to be poor. About 21 percent of
retired persons without pension income had incomes below the federal poverty
threshold, compared with 3 percent with pension income.

The probability that a person's income will begin to decrease at a certain
age fosters economic insecurity for many in the labor force.2 Employee
pension plans are a key component of the nation's multifaceted retirement
income security system--Social Security, pensions, personal savings, and to
the extent retired persons are willing and able to continue working, wages
and salaries--designed to address this issue. Participation in employee
pension plans provides a structured mechanism for persons in the labor force
to save for retirement. Income from such plans can be an important factor in
determining whether people can maintain their preretirement standard of
living after they retire.

Recognizing the importance of pension income to economic well-being, the
federal government uses the income tax system to encourage firms to sponsor
and employees to participate in such plans. At about $76 billion in 2000,
tax preferences for employer pension plans are the single largest tax
expenditure, exceeding tax subsidies for home mortgages and health
benefits.3 These tax benefits seek to raise private saving for employee
retirement and are structured to strike a balance between providing
incentives for firms to start and maintain tax-qualified plans and ensuring
that lower-income employees receive an equitable share of the subsidy.

The Revenue Act of 1921 initiated the preferential tax treatment of pension
plans, and the Revenue Act of 1926 and the Revenue Act of 1928 extended that
preferential treatment to a broad range of pension plans. As broad as the
tax benefit became in the 1920s, it initially had little influence on the
development of pension plans. Pension coverage spread quickly during the
1940s and 1950s, however, encouraged in part by the high marginal federal
tax rates enacted to finance World War II, wage controls during and after
the war that exempted pension benefits, and union efforts to negotiate
pension benefits through collective bargaining. Pension coverage continued
to increase during the 1960s and into the 1970s but stabilized and perhaps
decreased slightly during the 1980s.4

Employee pension plans are customarily classified into two major categories:
defined benefit plans and defined contribution plans. A defined benefit plan
promises a retirement benefit amount that is usually determined by salary
level and length of service. A defined contribution plan specifies
contributions to be made, but the benefits depend on investment performance.

In terms of coverage, defined benefit plans were the predominant type of
employee retirement program for many years. In 1975, the percentage of
pension plan participants who had a defined contribution plan as their
primary plan was only 29 percent, but this increased to 60 percent by 1992.5
The Pension and Welfare Benefits Administration estimates that, in 1996, 83
percent of all employees with pension coverage were in either a primary or
supplemental defined contribution plan.

About 53 percent of the employed labor force, 69 million persons, were
without a pension plan in 1998, a decrease of 5 percentage points over 10
years. (See fig. 1.) This change was the result of a growing proportion of
the labor force being employed by firms with pension plans. Between 1988 and
1998, the proportion of persons who were employed by firms that did not
sponsor a plan decreased from 46 percent to 39 percent. However, that
decrease was partially offset by a 2 percentage point increase in the number
of persons who lacked a plan, either because they were not eligible or
because they chose not to participate in their firm's plan.6 The source of
our data on pension coverage over the past 10 years, CPS supplements
conducted in March of each year, does not include the information needed to
estimate how many of these persons chose not to participate. However,
another survey, the CPS Contingent Work Supplement conducted in February
1999, indicates that about 28 percent of the labor force members who did not
participate in their firm's plan, or about 4 percent of the employed labor
force, chose not to participate in that plan.7

Note: The figure excludes persons in the labor force who were unemployed
(about 1.7 million persons in both years). See appendix II. The figure and
appendix exclude self-employed persons (about 14 million in 1998) because
the March CPS does not collect information on Keogh plans, a type of pension
plan available to the self-employed.

Source: March 1989 and 1999 CPS.

The decline between 1988 and 1998 in the proportion of the labor force
without pension coverage may result from a number of factors. Factors that
might affect a firm's decision to sponsor a plan include the strength of the
general economy and the relative priority that current and prospective
employees place on pension coverage compared with wages and other forms of
compensation. Firms might have been more likely to sponsor a plan in 1998
than in 1988, for example, because the economic expansion under way since
1991 had taken unemployment rates close to historical lows and a pension
plan can serve as a management tool to attract and retain needed employees.
Factors that might have increased employee interest in pension benefits
include concerns about Social Security's financial status and the increased
proximity of the members of the baby-boom generation to their retirement
age. In January 2000, for example, analysts at the Federal Reserve reported
that their surveys indicated that retirement-related reasons for saving have
consistently increased in importance since 1989.8 Among the possible reasons
given for the increased focus on retirement savings were the ongoing
discussions concerning the future of Social Security and the aging of the
baby-boom generation.

Despite the increase in coverage from sponsoring firms, about 14 percent of
the employed labor force, 18.3 million persons, were not included in their
firm's pension plan, either because they did not meet the plan's
qualification standards or because they chose not to participate in it.
Federal regulations and other laws permit firms to maintain tax-qualified
plans that exclude certain groups of employees. For example, a firm may
exclude employees with less than 1 year of job tenure from participating in
its tax-qualified pension plan.9

Our findings on the proportion of the labor force that lacks coverage are
generally consistent with those of other survey data sets. For example,
another BLS survey, a survey of public and private-sector establishments to
determine the incidence and characteristics of employee benefit plans, also
indicates that tens of millions of employees are not earning a pension
benefit at any given time. The BLS establishment surveys do indicate that
the proportion of employees who lack pension coverage may be smaller than is
indicated by CPS surveys. On the basis of establishment surveys conducted
between 1994 and 1997, BLS estimates that 42 percent of full-time and
part-time employees lack pension plans.10 BLS economists who analyzed the
two surveys stated that it is difficult to explain why CPS coverage
estimates tend to differ from those derived from establishment surveys. The
economists indicated that coverage estimates based on establishment surveys
may be more accurate because the CPS is a household survey and some
respondents may lack knowledge about their own benefits coverage or the
benefits coverage of the other household members for whom they responded.
The BLS economists also stated, however, that establishment surveys do not
include the detailed demographic and other information on employees who lack
pension coverage found in the CPS, limiting the analysis of the
characteristics of this population.

Coverage

About 85 percent of the persons in the labor force who worked and lacked
pension coverage had one or more of four characteristics: They had
relatively low income, did not work full time for the entire year, worked
for a relatively small firm, or were relatively young. In turn, each of
these factors is associated with the trade-off that current and prospective
employees might be willing or able to make between pension benefits and
other forms of compensation, such as wages and health benefits and the costs
and benefits to firms of sponsoring a pension plan for their employees. For
the most part, employees exhibited similar

characteristics,whether they were not covered because their firms did not
sponsor a plan or they chose not to participate in their firm's plan.

Likely to Lack Coverage

Employees with certain demographic or economic characteristics were more
likely to lack pensions. Figure 2 shows that employees with the following
characteristics were disproportionately more likely to lack pension
coverage: employees with incomes of less than $20,000 (81 percent),
employees who worked part time or part of the year (79 percent), employees
at firms with fewer than 25 employees (82 percent), and employees younger
than 30 (76 percent).11 Many employees without pension plans had two or more
of these characteristics. Together, for example, lower-income and small-firm
employees accounted for 74 percent of persons who had no pension plan.
Additionally, 82 percent of employees with incomes of less than $20,000 were
part-time or part-year employees or younger than 30 or both. Research has
provided a number of possible reasons for the association of these
characteristics with the lack of pension coverage.

Note: Reported income is individual income. Studies indicate that income may
be underreported in the CPS. See appendix II for the estimated number of
employees with each characteristic who lack pension coverage.

Source: March 1999 CPS.

Employees With Incomes of Less Than $20,000

While 34 percent of employees with incomes of $20,000 or more lacked pension
coverage, 81 percent with lower incomes lacked coverage. Our model
associated a 22 percentage point increase in the likelihood of being
pensionless for employees with incomes of less than $20,000 compared with
employees with incomes of more than $60,000. Income influences pension
coverage in a number of ways. First, lower-income employees have constraints
on the amount of money they can save because of their current consumption
needs. If a person's income barely covers a subsistence level of
consumption, regardless of age, he or she might not be able to afford to
save for retirement and might place a higher value on wages than on pension
benefits. As indicated by data on family net worth, lower-income families
(and hence workers) accumulate less savings. For example, in 1998, families
with incomes of $10,000 to $25,000 had a median net worth of $24,800,
whereas families with incomes greater than $100,000 had a median net worth
of $510,800.12

Persons with lower incomes also benefit less from the preferential treatment
afforded to pension benefits with respect to income taxes than do persons
with higher incomes. Lower-income employees typically have lower marginal
tax rates and therefore gain less from deferring taxes on pension plan
contributions and earnings than higher-income employees.13 Higher-income
persons also may have lower marginal tax rates during retirement, when taxes
would be due on pension plan distributions, than during their working years.
Lower-income individuals, in contrast, are less likely to receive this
benefit because their income is more likely to be taxed at the lowest rate
while they are working and in retirement. This could further reduce the
interest that lower-income workers have in pension plan coverage.

Beyond the obstacles to saving that lower-income persons face generally,
some analysts contend that government programs also might affect the
association of income and pension coverage to the extent that Social
Security, Supplemental Security Income (SSI), and other programs may curtail
the need for savings among lower-income persons.14 These analysts argue that
Social Security replaces a higher percentage of the lower-income person's
preretirement earnings. This means that lower-income persons can rely on
Social Security benefits to a greater extent to maintain their preretirement
standard of living than can higher-income persons. Additionally, SSI and
other needs-based programs have asset-based means tests to qualify for
benefits. They conclude that needs-based requirements could discourage some
lower-income workers from saving for retirement because the wealth
accumulation would reduce their chances of qualifying for the programs.

Part-Time and Part-Year Employees

While 40 percent of persons who were employed full time and year-round
lacked pension coverage, 79 percent of part-time and part-year employees
lacked coverage. Our model indicated that the likelihood of lacking coverage
is 20 percentage points higher for part-time workers than for full-time
workers. Persons working part time or part of the year are less likely to
have pension coverage than full-time or year-round employees, partly because
such employees may be less skilled or in more unstable jobs. In contrast,
increased full-time employment may spur pension sponsorship because of
firms' desire to reduce the turnover of skilled employees. Pension coverage
appears to reduce labor turnover, although this effect may be stronger at
larger firms than smaller firms.15 The reduction in turnover reduces the
firms' labor costs by lowering training costs, and it enhances productivity
by retaining skilled employees. Additionally, qualification standards for
pension participation permit firms that sponsor a plan to retain favorable
tax treatment, even if they exclude part-time and seasonal employees from
participation. Such exclusions might be significant for the 4.6 million
employees (3.5 percent of the labor force) who were involuntarily employed
part time in 1998; that is, they said they would prefer to work full time
but could not find full-time work.

Employees of Firms With Fewer Than 25 Employees

While 41 percent of persons employed by firms with 100 or more employees
lacked pension coverage, 82 percent of employees in firms with fewer than 25
employees lacked coverage. Our model associated a 14 percentage point
increase in the likelihood of lacking coverage for workers at firms with
fewer than 25 employees compared with workers at firms with 250 or more
employees. Several factors lead smaller firms to have lower rates of pension
coverage than larger firms. In a survey of small employers, instability of
revenues, employee preferences, and administrative burdens and costs were
frequently cited as reasons why small firms did not offer pension plans.16
Revenue uncertainty prevents many small employers from sponsoring pension
plans. About 19 percent of survey respondents said that revenues were too
uncertain for the firm to commit to a pension plan. Additionally, the
relationship between employee preferences and small-firm pension coverage is
exhibited by the type of workers a small firm employs. Smaller firms that
offer pension coverage tend to have employees with characteristics that are
associated with an increased interest in pension coverage--that is,
employees who have higher incomes, have more education, and stay with the
firm longer. Administrative burdens and costs also can hinder plan
sponsorship among small firms more than large firms because of economies of
scale. Administrative costs can be spread out over more employees in a
larger firm than in a smaller firm.

Employees Younger Than 30

While 42 percent of employees aged 30 to 59 lacked pension coverage, 76
percent of employees younger than 30 lacked coverage. At early stages of
their careers, some employees may prefer to not have pension coverage.
Younger employees generally choose to spend their incomes on immediate needs
such as paying child rearing expenses, paying college tuition, or financing
a mortgage rather than saving for retirement.17 Also, younger employees may
simply not feel the need to plan for retirement at the early stages of their
careers. As retirement nears, however, employees become more concerned about
planning for it and may increasingly select jobs with pension coverage.
Finally, employers are allowed to exclude employees from participating in a
plan during their first year of employment. This would affect younger
workers more than older workers.

Our analysis also found several other characteristics that are associated
with a lack of pension coverage. Education is correlated with income and the
type of job a person acquires. Having a relatively high educational level
makes a person more likely to work for a firm that offers coverage, and vice
versa.18 Job tenure also influences pension coverage; employees in
high-turnover industries may have less desire for coverage since they may be
unable to meet vesting requirements. In addition, employees in high-turnover
industries may prefer higher wages to pension benefits because their job
situation is more tenuous. Some industries and occupations lack pension
coverage to a greater extent than others. According to our analysis, for
example, in 1998, 58 percent of private-sector employees lacked pension
coverage, compared with 27 percent of public-sector employees. Public-sector
employees may have higher coverage rates than private-sector employees
because, among other reasons, many public-sector employees have
traditionally been excluded from Social Security coverage and more
public-sector employees are unionized.19 The collective action of union
members can push employers to offer pensions if the members of the union
want the coverage more than individuals in a nonunion setting.20

Race and ethnicity are also associated with a lack of pension coverage,
although this relationship is not well understood. According to our
analysis, blacks and non-Hispanic whites appear equally likely to lack
pension coverage, whereas Hispanics and Asians are more likely than
non-Hispanic whites to lack pension coverage. There are a limited number of
studies to explain these gaps in coverage, and more research is needed.21 In
addition, our analysis found associations between pension coverage and
industry of employment, occupation, marital status, and the spouse's pension
coverage status. However, as with race and ethnicity, more research is
needed to explain these differences.

Plan

About 4 percent of employees were offered coverage but chose not to
participate in a plan. We analyzed employees who were offered defined
contribution plan coverage and had no supplemental defined benefit coverage
to find the characteristics that were associated with their choosing not to
participate.22 Our analysis showed that employees who were more likely to
choose not to participate in defined contribution plans, such as 401(k)
plans, have some of the same characteristics as persons in the workforce who
lacked pension coverage. For example, employees who chose not to participate
in their firm's pension plans were more likely to have relatively low
income, be relatively young, and have less than a year of job tenure.23

Our analysis showed that unlike the persons with characteristics described
above, employees of larger firms (firms with 100 or more employees) compared
with employees of small firms were 11 percent more likely to choose not to
participate in their firms' defined contribution plans. A study by Fidelity
Investments of more than 5,400 defined contribution pension plans covering
5.2 million participants and using 1998 data also found that larger defined
contribution plans have higher nonparticipation rates than smaller plans.24
The study concluded that the work environment in small companies might allow
more opportunity for employees to interact with and be motivated by
management, which could lead to greater participation in a firm's pension
plan. Also, larger companies are more likely to have other retirement
programs, such as defined benefit plans, that automatically cover eligible
employees. Employees who are automatically enrolled in a defined benefit
plan might be less likely to choose to participate in a supplementary 401(k)
plan.

Our analysis also showed that employees were 12 percent more likely to
choose not to participate in their firm's defined contribution plan when the
firm did not provide matching contributions to the plan. The Fidelity
Investments study had similar results. It reported that a company match has
a very positive effect on participation rates, regardless of a plan's size.
In large plans, for example, a 1.5 percent effective match increases
participation by nearly 19 percentage points.

While it is not possible to predict how many persons currently in the labor
force will ultimately earn a pension benefit, our analysis indicates that in
the past, many workers failed to earn such a benefit.25 In 1998, for
example, 48 percent of retired persons (17.6 million of 36.6 million)
reported that they had no pension income of their own or from a spouse.26

Some of the characteristics we identified as associated with employees who
lacked pension coverage also applied to retired persons who lacked pension
income: They were more likely to lack high school diplomas, be Hispanic or
nonwhite, and have incomes below the federal poverty threshold. (See fig.
3.) For example, 21 percent of retired persons without pension income had
incomes below the federal poverty threshold, compared with 3 percent with
pension income. In addition, we found that women and single retired persons
were more likely to lack pension income. For example, 51 percent of retired
persons without pension income were single compared with 32 percent with
pension income.

Note: The number of persons with pension income = 19 million. The number of
persons without pension income = 17.6 million. See appendix II for the
estimated number of retirees with each characteristic who lacked pension
income.

Source: March 1999 CPS.

Retired persons without pension income were also less likely than retired
persons with pension income to have income from other sources, except for
SSI and other public assistance programs. (See fig. 4.) For example, persons
without pension income were less likely to have income from assets (51
percent) than those with pension income (78 percent). However, those without
pension income were more likely to rely on public assistance programs than
those with pension income.

Source: March 1999 CPS.

Although most of the 17.6 million retired persons who lacked pension income
did not have incomes below the federal poverty threshold, about 21 percent
of them (3.6 million) had incomes below the poverty threshold. The
characteristics associated with the lack of pension income shown in figure 3
were also highly associated with retired persons who had incomes below the
poverty threshold and who lacked pension income. For example, 71 percent of
retired persons who had incomes below the poverty threshold and lacked
pension income were single. Similarly, 69 percent were women, and 55 percent
lacked a high school diploma. (See table 1.)

                        Percent
 Characteristic
                        (N = 3.6 million)
 Single                 71%
 Women                  69
 No high school diploma 55
 Hispanic or not white  37

Source: March 1999 CPS.

In recent decades, federal efforts to expand private pension coverage have
concentrated on providing tax incentives for employers to sponsor and
employees to participate in pension plans. These efforts have been
relatively successful in fostering pension coverage among moderate-income
and high-income workers. Almost two-thirds of all workers with incomes
greater than $20,000 a year had pension coverage. Despite these longstanding
tax benefits and a sustained economic expansion featuring historically low
unemployment levels, more than half of the labor force in 1998--71 million
workers--continued to have no pension coverage. Further, this lack of
pension coverage is concentrated in certain segments of the labor force.
More than 80 percent of lower-income workers--those earning less than
$20,000 per year--continue to lack pension coverage.

We note that our analysis indicates only the extent to which labor force
participants do not have pension coverage at a point in time. Some of the
"pensionless" workers we identified will likely obtain a pension plan later
in their work careers. However, we also found that a sizable proportion of
current retired persons (48 percent) do not have pension income and that
they are more likely to have incomes below the federal poverty threshold
than those who receive pension income. This suggests that incentives or
interventions that rely on tax preferences to encourage voluntary pension
sponsorship and participation have not enabled certain segments of the labor
force, such as lower-income and part-time employees, to receive pension
benefits at retirement.

We provided Labor the opportunity to comment on the report. Labor provided
us with technical comments, which we incorporated where appropriate.

We are sending copies of this report to the Honorable Alexis M. Herman,
Secretary of Labor, and the appropriate congressional committees. Copies
will also be made available to others on request. Please call me at (202)
512-5491 or Charles A. Jeszeck at (202) 512-7036 if you or your staff have
any questions. Other major contributors to the report include Andrew
Davenport, Jeffrey S. Petersen, Donald J. Porteous, and John M. Schaefer.

Barbara D. Bovbjerg
Associate Director, Education, Workforce, and
Income Security Issues

Scope and Methodology

To obtain information on the characteristics of persons in the labor force
who do not have pension plans and the extent to which they are not covered,
we interviewed pension experts and reviewed relevant studies. Using
information from interviews and other studies, we identified the demographic
and other characteristics associated with pension coverage and analyzed
survey data. We used a logistic regression model to assess the influence
those characteristics appeared to have on pension coverage. We also used
survey data to provide current data for the relevant characteristics. We
used information from the studies that we reviewed and the pension experts
whom we interviewed to provide a context for the relevant characteristics.
Similarly, we reviewed studies and analyzed survey data to examine the
extent to which retired persons do not receive pension income and their
characteristics.

We used survey data from the Current Population Survey (CPS) because of its
large sample size, its inclusion of detailed information on the economic and
demographic characteristics of labor force participants, and the timeliness
of its data and because its collection frequency allows the opportunity to
show trends over time. The annual March CPS supplement is the official
source of income and poverty estimates for the United States and has been a
mainstay of income measurement in the United States since it began in 1947.
The March supplement includes household economic, demographic, and
employment data for the previous year. It contains labor force data on
101,000 persons aged 15 and older. The large sample size makes it possible
to estimate rates of pension coverage among categories of workers that are
based on various demographic characteristics with small sampling errors. The
sampling errors for the estimated percentages used in this report are less
than plus or minus 1 percentage point at the 95 percent confidence level. We
used annual March CPS supplements from 1989 and 1999 to provide trend data
concerning pension coverage. These supplements provide information
concerning pension plan sponsorship and participation for jobs held in the
year preceding the survey, although they do not contain information as to
why persons were not in their firm's plan--that is, whether they were
ineligible or chose to not participate.

For pension-related information not available in the March supplement, we
used other CPS supplements. We used the February 1999 CPS Contingent Work
Supplement for information about whether employees were ineligible or chose
not to participate in their current firm's pension plan. The most recent CPS
supplement specifically designed to provide comprehensive information
concerning pension coverage and participation was the April 1993 Survey of
Employee Benefits, which we used for our logistic regression models.
Comprehensive pension coverage and participation information was collected
in the 1996 Survey of Income and Program Participation; however, that
information has not yet been processed and released for public use.

Although widely used and a rich source of detailed data, CPS and other
surveys that are based on self-reported data are subject to several sources
of nonsampling error, including inability to get information about all
sample cases; difficulties of definition; differences in the interpretation
of questions; respondents' inability or unwillingness to provide correct
information; and errors made in collecting, recording, coding, and
processing data.

These nonsampling errors can influence the accuracy of information presented
in the report, although the magnitude of their effect is not known. For
March 1999, about 8 percent of persons in the sample did not respond to the
basic survey, which the Bureau of the Census adjusts for by using weighting
techniques. Additionally, some respondents do not provide answers to all the
questions in the survey, and the Bureau imputes responses based on data from
respondents with similar demographic characteristics. For example, about 14
percent of the data on pension inclusion and about 26 percent of the data on
earnings are imputed. These adjustments may not fully correct for missing or
incorrectly reported data. For example, much previous research has shown
that all sources of income have historically been underreported in the
CPS.27 Furthermore, when responses to pension questions were provided, more
than 20 percent of the respondents were persons in the household other than
the employee or the employee's spouse, and Bureau of Labor Statistics (BLS)
researchers question the ability of these persons to consistently provide
accurate information concerning the employee's pension coverage.

In addition, comparing data between supplements is difficult because of
differences in the wording of questions, the time periods the questions
cover, and the persons of whom the questions are asked. For example, the
March supplement asks pension questions concerning employment in the past
year, while the February supplement asks questions concerning current
employment.

We developed two models to examine the strength of the association between
various economic and work-related characteristics and the lack of pension
coverage or participation. In the first model, the pension coverage model,
we examined persons aged 21 to 54 who were not self-employed to assess the
factors that contribute to their lack of pension coverage. In the second
model, the pension choice model, we examined persons aged 21 to 54 who were
not self-employed who were offered defined contribution pension coverage and
assessed the factors that influence whether they choose to participate in a
plan. We used the April 1993 CPS Survey of Employee Benefits to test the
characteristics that appear to influence pension coverage and choice. The
models and their results are described in appendix III.

Analysis of CPS Pension Coverage Rates and Lack-of-Coverage Factors

Table 2 shows that excluding self-employed persons, almost 133 million
persons reported that they worked or looked for work at some time during
1998.28 About 69 million persons in the labor force were employed but lacked
pension coverage because their employers did not sponsor a plan or they were
not eligible for or chose not to participate in their employer's plan. An
additional 1.7 million persons lacked pension coverage because they were
unemployed throughout the year.

 Status                                                        1988  1998
 Employed labor force
 In a plan                                                     48.6  61.5
 Not in a plan
 Firm did not sponsor a plan                                   54.3  51.1
 Firm sponsored a plan, but person was not
                                                               14.1  18.3
 eligible or chose not to participate in it
 Total employed and not in a plan                              68.4  69.4
 Total employed labor force                                    117.0 130.9
 Unemployed-did not work but looked for work some time during
 the year                                                      1.7   1.7
 Total labor force                                             118.7 132.6

Note: Numbers are millions of persons.

Source: March 1989 and 1999 CPS.

Table 3 shows that a majority of the 69 million employees who lacked pension
coverage had specific characteristics. For example, employees with incomes
of less than $20,000 made up 40 percent of the employed labor force but 61
percent of those who lacked pension coverage. Similarly, persons who worked
part time or for part of a year made up 34 percent of the employed labor
force but 51 percent of those who lacked coverage.

                               Employed labor forcea Not in a pension plan
 Characteristic                Number    Percent     Number     Percent
 Total individual income
 Less than $20,000             52.3      40%         42.5       61%
 $20,000 or more               78.5      60          26.9       39
 Total                         130.9     100%        69.4       100%
 Work schedule
 Employed part time or part of
 the year                      44.9      34%         35.4       51%
 Employed full time and all
 year                          85.9      66          34.1       49
 Total                         130.9     100%        69.4       100%
 Firm size
 Fewer than 25 employees       29.2      22%         23.9       34%
 25−99 employees         17.8      14          11.2       16
 100 or more employees         83.9      64          34.3       49
 Total                         130.9     100%        69.4       100%
 Age in years
 16−29                   38.7      30%         29.5       42%
 30−59                   83.7      64          35.2       51
 60 or older                   8.5       6           4.8        7
 Total                         130.9     100%        69.4       100%

Note: Number of persons = millions. Numbers may not add because of rounding.

aExcludes 1.7 million persons in the labor force who were unemployed.

Source: March 1999 CPS.

At retirement, persons who lack pension income were more likely to have
incomes below the federal poverty threshold or to be single, women, Hispanic
or not white or to have less education compared with those with pension
income. (See table 4.) For example, 21 percent of retired persons without
pension income had incomes below the federal poverty threshold compared with
3 percent with pension income. Similarly, 23 percent without pension income
were Hispanic or not white compared with 11 percent with pension income.

                                         Pension income  No pension income
 Characteristic             Total number Number  Percent Number   Percent
 Total incomea
 Below poverty threshold    4.2          0.5     3%      3.6      21%
 Above poverty threshold    32.4         18.4    97      13.9     79
 Total                      36.6         19.0    100%    17.6     100%
 Marital status
 Single                     15.1         6.0     32%     9.0      51%
 Married                    21.5         12.9    68      8.5      49
 Total                      36.6         19.0    100%    17.6     100%
 Gender
 Male                       15.2         8.9     47%     6.3      36%
 Female                     21.3         10.0    53      11.3     64
 Total                      36.6         19.0    100%    17.6     100%
 Race or ethnicity
 White                      30.5         16.9    89%     13.6     77%
 Hispanic or not white      6.1          2.1     11      4.0      23
 Total                      36.6         19.0    100%    17.6     100%
 Education
 No high school diploma     11.1         4.0     21%     7.1      40%
 High school diploma or
 higher                     25.5         15.0    79      10.5     60
 Total                      36.6         19.0    100%    17.6     100%

Note: Number of persons = millions. Numbers may not add because of rounding.

aTotal income includes income of the spouse and other family members who
lived with the retired person.

Source: March 1999 CPS.

Summary of GAO's Regression Models

The purpose of the logistic regression models is to find associations
between pension coverage and the characteristics of employees and the firms
for which they work. The results of the regressions should not be
interpreted as direct causation between the lack of pension coverage and the
variables in a model. For the pension coverage model, we defined the outcome
variable as whether or not persons in the labor force were covered by an
employee pension plan. Closely following an analytical model presented by
pension experts, we categorized the predictor variables as employee- or
job-specific.29 Employee-specific variables assess the employees' demand for
a pension, attempting to capture their desire to have pension coverage or to
select themselves into jobs that have pension coverage. For example, as a
person's total income rises, he or she may be more likely to demand pension
coverage because of the tax advantages that pensions offer relative to wages
and other forms of compensation. Job-specific variables attempt to assess
the firm's willingness to supply a pension. For example, small firms lack
the economies of scale (as the number of employees increases, cost per
employee decreases) to start and administer pension plans. Therefore, small
firms may be less likely to provide coverage.

Table 5 presents the results of our pension coverage model. The model
indicates that the statistically significant employee characteristics
associated with the lack of coverage were income, part-time status, age,
tenure, race, education, firm size, union status, the spouse's pension
coverage, the spouse's employment status, industry of employment, and type
of occupation.30 The amount of the difference in pension coverage that is
attributable to the difference between the "reference" variable and the
other variables in the category is shown in the marginal effect column.
Marginal effect shows the likelihood of not having pension coverage if an
individual moved from the reference variable to another variable within the
category. For example, the table shows that the marginal effect of being a
part-time worker is 20.1 percentage points. Evaluated at the mean
characteristic across all workers in the sample, the results suggest that if
a full-time employee switched to a part-time job, the likelihood of this
employee lacking a pension would increase by 20.1 percentage points, if all

the person's other characteristics that were included in the model held
constant.

                            Percentage
     Characteristic          without
                            coverage        Logit      Standard   Marginal
                                          estimate      error     effect
 Income
 Less than $20,000       49.9%          1.389        0.131      22.2%   **
 $20,000 to $39,999      20.8           0.592        0.123      13.6    **
 $40,000 to $59,999      11.0           0.221        0.131      5.5
 More than $59,999       9.4            Referencea
 Work status
 Part-time               55.3           1.040        0.083      20.1    **
 Full-time               22.8           Referencea
 Age in years
 21-29                   35.4           0.146        0.072      3.6     *
 30-39                   26.4           0.019        0.062      0.5
 40-54                   33.5           Referencea
 Tenure
 1 year or less          51.3           0.681        0.086      15.2    **
 More than 1 year        24.0           Referencea
 Race or ethnicity
 Black                   23.6           0.148        0.099      3.7
 Hispanic                42.1           0.445        0.100      10.6    **
 Asian                   29.6           0.380        0.177      9.2     *
 Other                   28.5           0.273        0.347      6.7
 White                   25.0           Referencea
 Gender
 Male                    24.7           0.063        0.066      1.6
 Female                  27.9           Referencea
 Marital status and
 spouse's work status and
 coverage
 Married, spouse works
 but has no coverage     32.7           0.219        0.076      5.4     **
 Married, spouse does not
 work                    25.1           0.010        0.083      0.2
 Married, spouse works
 and has coverage        20.4           -0.243       0.067      -6.0    **
 Single                  29.7           Referencea
 Education
 No high school diploma  52.3           0.826        0.120      17.5    **
 High school diploma     31.4           0.441        0.086      10.5    **
 Some college            24.7           0.163        0.084      4.0
 College diploma         14.4           Referencea
 Firm size
 Fewer than 25 employees 75.8           2.950        0.073      13.9    **
 25-99 employees         45.2           1.927        0.072      21.4    **
 100-249 employees       22.5           0.904        0.093      18.5    **
 250 employees or more   10.3           Referencea
 Union status
 Not a union member      31.6           0.995        0.088      19.6    **
 Union member            7.4            Referencea
 Industry
 Agriculture, forestry,
 and fisheries           69.3           0.796        0.307      17.1    **
 Mining                  21.0           0.808        0.295      17.2    **
 Construction            46.6           0.711        0.143      15.7    **
 Manufacturing,
 nondurable goods        19.7           0.283        0.127      6.9     *
 Transportation          22.5           0.811        0.154      17.3    **
 Communications          12.2           0.586        0.253      13.5    *
 Utilities               4.1            -1.173       0.360      -21.2   **
 Wholesale trade         29.7           0.204        0.147      5.1
 Retail trade            47.1           0.914        0.121      18.7    **
 Finance, insurance, and
 real estate             17.0           -0.036       0.145      -0.9
 Business and repair     46.1           0.963        0.142      19.3    **
 Personal services,
 including private       73.0           1.527        0.199      22.4    **
 household
 Entertainment           44.3           0.917        0.237      18.7    **
 Hospital                4.9            -0.705       0.210      -15.6   **
 Medical, except hospital38.5           0.093        0.154      2.3
 Education               13.3           0.179        0.148      4.4
 Social services         45.1           0.821        0.190      17.4    **
 Other professional      31.3           0.325        0.153      7.9     *
 Public administration   3.8            -1.027       0.235      -19.9   **
 Manufacturing, durable
 goods                   13.4           Referencea
 Occupation
 Executive,
 administrative, and     21.8           0.230        0.112      5.7     *
 managerial
 Technician              16.4           -0.224       0.165      -5.5
 Sales                   36.2           0.191        0.130      4.7
 Administrative support  20.6           -0.309       0.118      -7.5    **
 Private household       95.8           0.916        0.728      18.7
 Protective service      11.3           0.482        0.271      11.4
 Other service           50.9           0.461        0.132      10.9    **
 Precision production    28.5           0.057        0.137      1.4
 Machine operator        25.2           0.132        0.156      3.3
 Transportation          31.5           0.212        0.164      5.2
 Handler                 37.8           0.337        0.173      8.2
 Farming, fishing, and
 forestry                73.9           0.700        0.328      15.5    *
 Professional specialty  14.0           Referencea

aReference = identifies the reference variable.

** = statistical significance at the .01 level.

* = statistical significance at the .05 level.

Source: GAO analysis of April 1993 CPS.

We also developed our pension choice model by closely following the
analytical models pension experts use.31 This model assesses several factors
that could influence an employee's decision to participate in a

defined contribution pension plan.32 For this model, the outcome variable is
whether persons in the labor force choose to participate in a plan of this
type of plan when the firm offers it. Because the factors influencing
participation and coverage are similar, we generally used the same predictor
variables in the pension choice model. We added a variable to assess whether
the presence of a firm's matching contribution is associated with
participation. Employees who had defined-benefit coverage or were offered
multiple defined-contribution plans were excluded from the model. These
workers were excluded from the model in order to focus our analysis on
workers with only the choice to participate in a 401(k) plan, with the
401(k) plan being the only plan offered by the firm.

Table 6 shows the characteristics associated with persons in the labor force
who chose not to participate in a plan. Lack of a matching contribution by
the firm, lower earnings, and shorter tenure were associated with an
increased likelihood of choosing not to participate. Younger persons were
less likely to participate compared with persons of middle age. Employees in
firms with fewer than 100 employees were more likely to participate than
employees in firms with 100 or more employees. Black employees were less
likely to participate compared with white employees.

                             Percent
     Characteristic          without
                            coverage        Logit     Standard    Marginal
                                          estimate     error      impact
 Income
 Less than $20,000        37.3%         1.490        0.312      22.4%   **
 $20,000 to $39,999       16.4          0.604        0.288      13.8    *
 $40,000 to $59,999       8.3           -0.061       0.310      -1.5
 More than $59,999        7.6           Referencea
 Employer's matching
 contribution
 None                     19.7          0.523        0.165      12.2    **
 Some                     16.8          Referencea
 Work status
 Part-time                27.2          0.061        0.255      1.5
 Full-time                16.8          Referencea
 Age in years
 21-29                    26.4          0.667        0.183      14.9    **
 30-39                    16.9          0.374        0.166      9.0     *
 40-54                    12.1          Referencea
 Tenure
 1 year or less           47.4          1.684        0.203      22.2    **
 More than 1 year         14.9          Referencea
 Race or ethnicity
 Black                    29.9          0.603        0.239      13.8    *
 Hispanic                 26.5          0.334        0.293      8.1
 Asian                    6.49          -1.076       0.601      -20.4
 White                    16.37         Referencea
 Gender
 Male                     14.9          0.177        0.162      4.4
 Female                   20.5          Referencea
 Marital status and
 spouse's work status and
 coverage
 Married, spouse works
 but has no coverage      20.3          0.456        0.191      10.8    *
 Married, spouse does not
 work                     21.6          0.297        0.212      7.3
 Married, spouse works
 and has coverage         16.5          -0.359       0.168      -8.7    *
 Single                   12.7          Referencea
 Education
 No high school diploma   23.7          0.110        0.362      2.7
 High school diploma      19.0          0.310        0.216      7.6     *
 Some college             21.2          0.494        0.192      11.6    *
 College diploma          12.0          Referencea
 Firm size
 100 or more employees    18.3          0.455        0.203      10.8    *
 Fewer than 100 employees 13.2          Referencea
 Union status
 Not a union member       14.5          0.178        0.230      4.4
 Union member             17.8          Referencea
 Industry
 Construction             14.5          0.316        0.492      7.7
 Manufacturing,
 nondurable goods         12.6          -0.139       0.291      -3.5
 Transportation,
 communications, and      18.6          0.214        0.330      5.3
 utilities
 Wholesale trade          11.7          -0.110       0.347      -2.7
 Retail trade             22.0          0.495        0.263      11.7
 Finance, insurance, and
 real estate              20.4          0.521        0.267      12.2
 Business and repair      16.9          0.101        0.320      2.5
 Hospital                 35.0          1.263        0.292      21.7    **
 Medical, except hospital 19.1          0.051        0.423      1.3
 Education                12.1          -0.079       0.390      -2.0
 Other professional       11.6          0.078        0.380      2.0
 Public administration    19.5          0.383        0.352      9.2
 Manufacturing, durable
 goods                    14.5          Referencea
 Occupation
 Executive,
 administrative, and      9.8           -0.635       0.272      -14.4   *
 managerial
 Technician               14.9          -0.637       0.336      -14.4
 Sales                    13.4          -0.734       0.324      -16.1
 Administrative support   25.2          -0.095       0.264      -2.4
 Private household and
 other service            30.2          -0.023       0.384      -0.6
 Precision production     16.9          -0.037       0.326      -0.9
 Machine operator         19.7          0.018        0.355      0.4
 Transportation           18.8          -0.382       0.459      -9.2
 Handler                  27.1          -0.489       0.436      -11.5
 Professional specialty   15.8          Referencea

aReference = reference variable.

** = statistical significance at the .01 level.

* = statistical significance at the .05 level.

Source: GAO analysis of April 1993 CPS.

(207081)

Table 1: Characteristics of Retired Persons Who Had Incomes Below
the Poverty Threshold and Lacked Pension Income, 1998 22

Table 2: Pension Status of the Wage and Salary Labor Force, 1988
and 1998 27

Table 3: Major Characteristics of Employees Not in a Pension
Plan, 1998 28

Table 4: Characteristics of Retired Persons With No Pension Income
of Their Own or From a Spouse, 1998 29

Table 5: Characteristics Associated With Persons in the Labor Force
Who Lacked Pension Coverage, 1993 31

Table 6: Characteristics Associated With Persons in the Labor Force
Who Chose Not to Participate in a Pension Plan of the 401(k)
Type, 1993 34

Figure 1: The Proportion of the Employed Labor Force That Lacked Pension
Coverage, 1988 and 1998 8

Figure 2: The Proportion of Employees With Selected Characteristics
Who Did Not Have a Pension Plan, 1998 12

Figure 3: Percentage of Retired Persons With and Without Pension
Income Who Had Selected Characteristics, 1998 20

Figure 4: Sources of Income for Retired Persons With and Without
Pension Income, 1998 21
  

1. While Social Security coverage is mandatory and nearly universal, it is
not designed to provide adequate retirement income by itself. Pensions,
personal savings, or earnings from work are usually needed to fill the gap
between Social Security benefits and an adequate retirement income.

2. BLS includes in the labor force all members of the civilian,
noninstitutional population who are aged 16 or older and either working
(employed) or actively looking for work (unemployed).

3. The Joint Committee on Taxation has also estimated that tax expenditures
were $5 billion for Keogh plans for the self-employed and $12.2 billion for
individual retirement plans. Joint Committee on Taxation, Estimates of
Federal Tax Expenditures for Fiscal Years 2000-2004, JCS-13-99 (Washington,
D.C.: Dec. 22, 1999).

4. Department of Labor, Pension Coverage Issues for the '90s (Washington,
D.C.: 1994), and Pension and Health Benefits of American Workers
(Washington, D.C.: May 1994).

5. 401(k) Pension Plans: Many Take Advantage of Opportunity to Ensure
Adequate Retirement Income (GAO/HEHS-96-176, Aug. 2, 1996).

6. While defined benefit plans generally enroll qualified employees
automatically, defined contribution plans, such as 401(k) plans, are
generally voluntary and employees must choose to participate in them.

7. Estimates based on the February and March CPS 1999 supplements are not
strictly comparable. The February CPS supplement defined the labor force in
terms of current employment, while the March CPS supplement defined it in
terms of employment during the year preceding the survey. According to the
Pension and Welfare Benefits Administration, for example, the March
supplement elicits a higher number of part-year employees than the February
supplement.

8. Arthur B. Kennickell, Martha Starr-McCluer, and Brian J. Surette, "Recent
Changes in U.S. Family Finances: Results from the 1998 Survey of Consumer
Finances," Federal Reserve Bulletin (Washington, D.C.: Jan. 2000).

9. The Internal Revenue Code and associated regulations establish rules that
pension plans must meet to qualify for favorable tax treatment, including
minimum length-of-service and age conditions for plan participation.
Generally, plans must consider employees eligible if they have 1 year of
service and are at least 21. A year of service is defined as a 12-month
period in which the employee has at least 1,000 hours of service.

10. For a variety of reasons, pension coverage rates calculated on the basis
of establishment surveys and the CPS are not directly comparable. For
example, surveys of employer establishments exclude some employee groups
that are included in CPS surveys. However, BLS economists found that after
they adjusted for these differences, CPS estimates of participation in
pension plans are considerably lower than estimates derived from
establishment surveys. Diane E. Herz, Joseph R. Meisenheimer II, and Harriet
G. Weinstein, "Health and Retirement Benefits: Data from Two BLS Surveys,"
Monthly Labor Review (Mar. 2000).

11. We identified major characteristics associated with the lack of pension
coverage using a model described in appendix III.

12. Arthur B. Kennickell, Martha Starr-Mcluer, and Brian J. Surette, "Recent
Changes in U.S. Family Finances: Results from the 1998 Survey of Consumer
Finances," Federal Reserve Bulletin (Jan. 2000).

13. In 1999, the tax code had five marginal tax rates. The lowest rate was
15 percent for taxable income below $25,750 ($43,050 for married couples
filing jointly), and the highest rate was 39.6 percent for taxable income
higher than $283,150 for single persons and married couples.

14. R. Glenn Hubbard, Jonathan Skinner, and Stephen P. Zeldes,
"Precautionary Saving and Social Insurance," Journal of Political Economy,
Vol. 103, No. 2 (1995), pp. 360-99.

15. William E. Even and David A. Macpherson, "Employer Size and Labor
Turnover: The Role of Pensions," Industrial and Labor Relations Review, Vol.
49, No. 4 (July 1996), pp. 707-28.

16. Paul Yakoboski, Pamela Ostuw, and Bill Pierron, The 1999 Small Employer
Retirement Survey: Building a Better Mousetrap Is Not Enough, Employee
Benefit Research Institute Issue Brief 212 (Washington, D.C.: Aug. 1999).

17. The theoretical basis for such preferences is found in a life-cycle
savings and consumption model. According to this model, expenditures
typically exceed earned income during the early part of a person's working
years because of more immediate consumption needs. This tends to discourage
saving for retirement. Later in life, consumption needs are more likely to
be less than earned income, and households can pay off past debts and save
for retirement. At retirement, income decreases, consumption needs again
exceed earned income, and households use their savings. For a full
description of the life-cycle savings and consumption model, see Olivia S.
Mitchell and James F. Moore, Retirement Wealth Accumulation and
Decumulation: New Developments and Outstanding Opportunities (Philadelphia:
The Pension Research Council, Apr. 1997).

18. William E. Even and David A. Macpherson, "The Changing Distribution of
Pension Coverage," Industrial Relations, Vol. 39, No. 2 (Apr. 2000), pp.
199-227.

19. Olivia S. Mitchell and Anna M. Rappaport, "Innovations and Trends in
Pension Plan Coverage, Type, and Design," The Future of Pensions in the
United States (Philadelphia: The Pension Research Council, 1993).

20. Richard B. Freeman and James L. Medoff, What Do Unions Do? (New York:
Basic Books, 1984).

21. William E. Even and David A. Macpherson, "Racial and Ethnic Differences
in Pension Coverage and Benefit Levels," working paper. Hypotheses for the
differences in coverage rates among racial and ethnic groups include
differences in life expectancy and family structure.

22. Annual March and February versions of the CPS do not provide
comprehensive information on employees who have the choice to be in a plan
but choose not to participate in it. For example, the February 1999 CPS
lacks data on firm size. The most current data source with such information
is the April 1993 CPS Survey of Employee Benefits.

23. See appendix III for the complete results of our analysis.

24. Fidelity Investments, Building Futures: How American Companies Are
Helping Their Employees Retire. (n.p.: n.d.).

25. Persons who are retired now had work lives that spanned the past three
or four decades, which included the passage of the Employee Retirement
Income Security Act of 1974, the rise of 401(k) plans, and other major
pension developments. Thus, we cannot infer from a profile of current
retirees what the experience will be for current workers who are still many
years from retirement.

26. Using March CPS data, we categorized as "retired" persons aged 65 or
older not working full time and persons younger than 65 who gave "retired"
as the reason for not working during all or part of the year. We defined
pension income as including regular payments from annuities, individual
retirement accounts, Keogh or 401(k) accounts, veterans' pensions, or
survivor benefits. Annuities are insurance products that provide a stream of
payments for a preestablished period of time in return for a premium
payment.

27. Patrick J. Purcell, Pension Issues: Lump-Sum Distributions and
Retirement Income Security (Washington, D.C.: Congressional Research
Service, Mar. 14, 2000).

28. This figure excludes 14 million self-employed persons because the March
CPS does not include information on Keogh plans, a type of pension plan
available to the self-employed.

29. William E. Even and David A. Macpherson, "The Gender Gap in Pension and
Wages," Review of Economics and Statistics, Vol. 72, No. 2 (May 1990), pp.
259-62.

30. Variables were considered statistically significant at the .05 level.

31. Richard P. Hinz and John A. Turner, "Pension Coverage Initiatives: Why
Don't Workers Participate?" pp. 17-37, and R. L. Clark and S. J. Schieber,
"Factors Affecting Participation Rates and Contribution Levels in 401(k)
Plans," pp. 69-97, in Olivia Mitchell and Sylvester J. Schieber (eds.),
Living with Defined-Contribution Pensions (Philadelphia: The Pension
Research Council, 1998).

32. We considered individuals as participating in a plan of the 401(k) type
if they responded "yes" to the following question: "Some retirement plans
allow workers to make tax-deferred contributions to the plan. For example,
you might choose to have part of your salary deposited into a retirement
savings account, and then you do not pay income taxes on this money until
you take it out or retire. These plans are called by different names,
including 401(k) plans, pre-tax plans, salary reduction plans, and 403(b)
plans. Do you participate in a plan like this?" We classified individuals as
covered by a 401(k)-type plan if they said their employer offered such a
plan, whether or not they participated in it.
*** End of document. ***